geohabnet - Geographical Risk Analysis Based on Habitat Connectivity

The 'geohabnet' package is designed to perform a geographically or spatially explicit risk analysis of habitat connectivity. Xing et al (2021) <doi:10.1093/biosci/biaa067> proposed the concept of cropland connectivity as a risk factor for plant pathogen or pest invasions. As the functions in 'geohabnet' were initially developed thinking on cropland connectivity, users are recommended to first be familiar with the concept by looking at the Xing et al paper. In a nutshell, a habitat connectivity analysis combines information from maps of host density, estimates the relative likelihood of pathogen movement between habitat locations in the area of interest, and applies network analysis to calculate the connectivity of habitat locations. The functions of 'geohabnet' are built to conduct a habitat connectivity analysis relying on geographic parameters (spatial resolution and spatial extent), dispersal parameters (in two commonly used dispersal kernels: inverse power law and negative exponential models), and network parameters (link weight thresholds and network metrics). The functionality and main extensions provided by the functions in 'geohabnet' to habitat connectivity analysis are a) Capability to easily calculate the connectivity of locations in a landscape using a single function, such as sensitivity_analysis() or msean(). b) As backbone datasets, the 'geohabnet' package supports the use of two publicly available global datasets to calculate cropland density. The backbone datasets in the 'geohabnet' package include crop distribution maps from Monfreda, C., N. Ramankutty, and J. A. Foley (2008) <doi:10.1029/2007gb002947> "Farming the planet: 2. Geographic distribution of crop areas, yields, physiological types, and net primary production in the year 2000, Global Biogeochem. Cycles, 22, GB1022" and International Food Policy Research Institute (2019) <doi:10.7910/DVN/PRFF8V> "Global Spatially-Disaggregated Crop Production Statistics Data for 2010 Version 2.0, Harvard Dataverse, V4". Users can also provide any other geographic dataset that represents host density. c) Because the 'geohabnet' package allows R users to provide maps of host density (as originally in Xing et al (2021)), host landscape density (representing the geographic distribution of either crops or wild species), or habitat distribution (such as host landscape density adjusted by climate suitability) as inputs, we propose the term habitat connectivity. d) The 'geohabnet' package allows R users to customize parameter values in the habitat connectivity analysis, facilitating context-specific (pathogen- or pest-specific) analyses. e) The 'geohabnet' package allows users to automatically visualize maps of the habitat connectivity of locations resulting from a sensitivity analysis across all customized parameter combinations. The primary functions are msean() and sensitivity analysis(). Most functions in 'geohabnet' provide three main outcomes: i) A map of mean habitat connectivity across parameters selected by the user, ii) a map of variance of habitat connectivity across the selected parameters, and iii) a map of the difference between the ranks of habitat connectivity and habitat density. Each function can be used to generate these maps as 'final' outcomes. Each function can also provide intermediate outcomes, such as the adjacency matrices built to perform the analysis, which can be used in other network analysis. Refer to article at <https://garrettlab.github.io/HabitatConnectivity/articles/analysis.html> to see examples of each function and how to access each of these outcome types. To change parameter values, the file called 'parameters.yaml' stores the parameters and their values, can be accessed using 'get_parameters()' and set new parameter values with 'set_parameters()'. Users can modify up to ten parameters.